Tuesday, June 8, 2021

How do you find loops?

Ok, a little explanation as this happened to me a couple of times already. 3 or 4 network switches of different type, usually a Cisco in the mix. All at almost full capacity. Network rack looks like spaghetti cat5 vomit And access to these switches is undocumented. To put it in perspective, today i had to deploy wifi to a new location and discovered 2 Unifi 48 port switches. There were also 3 Cisco (2 were off) and a tp-link. The tp-link was fully used. I did not find a Unifi controller in any of the servers a f no one knew who installed them, so i proceed to boot up the cloud key that was just purchased. I logged into it and i accepted the adoption.....boom network went down. Even with static i could not login to the servers...until i tested using fqdn account instead of pre 2k....(i had the wrong pre2k) Now logging in, i realize that the problem is the firewall, so i connect directly to the sonicwal and my coworker connects to the modem. Modem works on IPv6 so internet is up, but firewall is not responding. I restart the firewall, same problem, so i remember a problem with a Sonicwal once when X0 decided to stop working and only X3 would work....but that was not it (or it is possible it was) I connect the firewall from the tp-link to the Cisco and it works, so now i am confused.... First thing i thought was that someone set VLAN on the Unifi switches using ssh (i have seen that sadly), but that was not it.... I never found the IP for the tp-link switch...as the Mac address sticker must be on top or bottom and it is not visible thanks to the other switches, but the Cisco was turning off and on 5 ports due to STP. I used what it was showing for CDP neighbor (really old iOS so even that only showed the ports) and the uplink in the Unifi to draw a diagram and found 2 loops....i found another by checking connection manually but i still think there is one more. Besides checking manually, i don't know of another way. What i think that happened is that the switch restart in the Unifi after a firmware upgrade got the stp running in the Cisco. I cannot prove it, but it seems sonicwal must have some implementation as well because this is the second time a Sonicwal locks up when a Cisco switch triggers stp to stop a loop. Cisco is very quick at that. I had it activated in a 8 port switch with just 1 cable connected......twice on different switches different places. People tell me i am crazy, but removing both switches fixed all the problems. Only way to avoid them is better documentation, which we are working, but i am wondering if there is another technique to get this done better. I am just lucky it was me on site, and not someone else, or if i was working remotely. So, how do you check loops? Any other suggestions? Magic? Sacrifices to the Ethernet gods?



No comments:

Post a Comment